01 / FROM IDEA TO OUTPUT

A workflow you can follow

01

Build one scene before a whole film

Create the project, choose its output size and place one title with one visual. Keep fonts and images available locally. A small composition lets you diagnose rendering issues without searching through a long timeline.

02

Make every state seekable

The renderer asks for a particular time; the animation must reach that state without waiting for the previous frame. Derive the camera, text and data values from that time. Use the frame lab below to see the difference between scrubbing a timeline and merely letting a page run.

03

Bind layers to the same clock

Assign each clip a start and duration. Make audio cues explicit rather than triggering them from a hover or a timeout. Preview transitions at normal speed, then seek around their boundaries to catch missing or doubled frames.

04

Export early, then scale the design

Render a short clip before adding particles, shaders or a second aspect ratio. Keep a known-good project version. When a change breaks export, compare the added asset or timing behavior against that baseline.

02 / START HERE

Make the brief your own

This is an original Hermes example. Replace the inputs and delivery requirements before sending it to your agent. It is not the showcase creator’s original prompt or a promise of the same result.

In this HyperFrames project, build a 12-second explanation of a fictional queue. Use three jobs labeled A, B and C and one worker. Explain waiting → processing → complete. 1920×1080, 30 fps. Put every state on a seekable timeline and label all values as example data. Load fonts before capture. First export 3 seconds; then deliver the complete MP4 and source.
INPUT

Real assets, the mechanism to explain and verified numbers.

OUTPUT

HTML compositions · Specify dimensions, duration and files.

REVIEW

Stills → short test → full output → independent playback or interaction.

03 / READ THE WORK

Scrub a frame. Understand the timeline.

INTERACTIVE STUDYHERMES STUDY · EXAMPLE DATA

ONE CLOCK. EVERY FRAME.

Same time. Same scene.

01 / COLLECTOne clear idea.
Brief.md
02 / MAKEShow the proof.
Scene.html
03 / FINISHReady to share.
✓
Film.mp4
04.00 s
120 / 359
12 s · 30 fps · Reveal

Seek to any frame, then return to it: the scene and readout match. Change the curve to compare the pacing. This Hermes teaching study explains a timeline; it is not a tool-generated result or performance benchmark.

04 / REVIEW & REVISE

When the result feels wrong

The first captured frame is blank

Verify asset loading and the composition’s initial state. Test an explicit seek to time zero before the full render.

Particles jump when you scrub backward

Use seeded randomness and time-derived positions. Accumulating positions on every browser tick makes a scene dependent on playback history.

Audio lands late

Check the encoded file at the visual impact and compare the waveform position. Fix the cue on the shared timeline, not with a separate playback delay.

Questions before you begin

Can I reuse an existing webpage?

It can supply assets and layout ideas. A reliable render still needs known duration, dimensions, loaded assets and seekable animation; pointer-triggered behavior needs a scripted equivalent.

What does the preview demonstrate?

Actova is the Motion Director author’s example built around HyperFrames. The local frame lab is a Hermes teaching demo; it demonstrates a shared clock and is not a benchmark of HyperFrames rendering.

Sources & display notes

The original projects are authoritative for capabilities and setup. The production recipes, sample briefs and review advice here are Hermes editorial work. A recommendation is not a claim of exhaustive tool testing.

HyperFrames repository & quickstart ↗Official documentation ↗Actova source project ↗